Uranium Transport in FCl-Bearing Fluids and Hydrothermal Upgrading of UCu Ores in IOCG Deposits

Uranium mineralization is commonly accompanied by enrichment of fluorite and other F-bearing minerals, leading to the hypothesis that fluoride may play a key role in the hydrothermal transport of U. In this paper, we review the thermodynamics of U(IV) and U(VI) complexing in chlorideand fluoride-bearing hydrothermal fluids and perform mineral solubility and reactive transport calculations to assess equilibrium controls on the association of F and U. Calculations of uraninite and U3O8(s) solubility in acidic F-rich (Cl : F = 100 [ppm-based]) hydrothermal fluids at 25–450 °C, 600 bar, show that U(IV)-F complexes (reducing conditions) and uranyl-F complexes (oxidizing conditions) predominate at low temperature (T < ~200°C), while above ~250°C, chloride complexes predominate in acidic solutions. In the case of uraninite, solubility is predicted to decrease dramatically as U(IV)Cl2 2+ becomes the predominant U species at T > 260°C. In contrast, the solubility of U3O8(s) increases with increasing temperatures. We evaluated the potential of low-temperature fluids to upgrade U and F concentrations in magnetite-chalcopyrite ores. In our model, an oxidized (hematite-rich) granite is the primary source of F and has elevated U concentration. Hydrothermal fluids (15wt.% NaCl equiv.) equilibrated with this granite at 200°C react with low-grade magnetite-chalcopyrite ores. The results show that extensive alteration by these oxidized fluids is an effective mechanism for forming ore-grade Cu-U mineralization, which is accompanied by the coenrichment of fluorite. Fluorite concentrations are continuously upgraded at the magnetite-hematite transformation boundary and in the hematite ores with increasing fluid : rock (F/R) ratio. Overall, the model indicates that the coenrichment of F and U in IOCG ores reflects mainly the source of the oreforming fluids, rather than an active role of F in controlling the metal endowment of these deposits. Our calculations also show that the common geochemical features of hematite-dominated IOCG deposits can be related to a two-phase process, whereby a magnetite-hematite-rich orebody (formed via a number of processes/tectonic settings) is enriched in Cu±U and F during a second stage (low temperature, oxidized) of hydrothermal circulation.


Introduction
1.1.The U-F Association.Many hydrothermal uranium (U) ores are accompanied by high concentrations of fluorine (F) in the form of fluorite and/or other F-bearing minerals such as fluorapatite [1][2][3][4].Such U-mineralizing systems include IOCG, orogenic U deposits, and volcanogenic and vein-type deposits hosted in felsic rocks [2,[5][6][7].Examples of deposits in which both U and F are enriched are listed in Table 1.
The enrichment of fluorine is an important feature of many IOCG-mineralizing systems, including Olympic Dam (OD) [6,8] and Ernest Henry in Australia [9,10]; Salobo [11] and Igarape Bahia [8] in the Carajás district, Brazil; and Lala [12,13] in southwest China.This led to the hypothesis that fluorine may play a key role in the leaching and transport of metals for IOCG and in controlling the element association found in this class [3,11].For example, in the Olympic Dam deposit, which represents the world's largest U resource [14][15][16][17], the mined ores contain about 2.5 wt.% fluorite, which adds up to 10 8 t of F from fluorite alone [3].The Ernest Henry in Cloncurry district, Australia, has a total reserve of 167 Mt ores with up to 50 ppm U, and fluorite is one of the main accessory phases closely associated with Cu-U-Au mineralization [6,10].Similarly, fluorite is enriched in the Chabiron et al. [22]; Hu et al. [7] Streltsovka caldera, Russia 0.127 0.18-0.2%U 1.4-2.7%F in rhyolite Felsic volcanic rocks/ rhyolites Chabiron et al. [22]; Cuney [27] Xiazhuang, China 0.003-0.010.1-0.3%Ore-forming fluids have elevated F concentration Granite T = 150 -250 °C Hu et al. [7] 2 Geofluids stage III ore assemblages together with Cu-Mo sulfides in the Lala deposit [18].Fluorite is usually enriched in volcanogenic and vein-type U deposits that are related to igneous rocks [4,[19][20][21].For example, fluorite veins are common at the Streltsovka caldera U deposit in Russia.The deposit is hosted in F-rich rhyolite (containing 1.4-2.7 wt.% F) over a granitic basement, and both rhyolite and granites are thought to be the source of F and U mineralization [22].Many vein-type U deposits found in South China (e.g., Xiangshan, Xiazhuang, Hechaokeng, and Wuqilin) are hosted by felsic intrusive or volcanic rocks and formed from fluids (100-250 °C) that were enriched in F [7].A close association between fluorite and U has also been reported in some orogenic U deposits, such as in the Mount Isa Inlier, Australia [2].Again, felsic rocks represent the source of U and F, with F assumed to facilitate U hydrothermal transport.
In summary, these studies reveal a clear association between F and U in the ores and sometimes in the likely source region of fluids/metals.Many studies suggest that this association is significant for understanding ore formation, with explanations focusing mainly on two aspects: (i) F helps extract U or (ii) the F-rich nature of the fluids reflects the source of the metals.In the case of felsic melt-aqueousphase separation relevant for magmatic hydrothermal deposits (e.g., Rijssing alaskites from Namibia [23] and Kvanefjeld deposit in the Ilimaussaq intrusion [24]), hightemperature experiments show that the partitioning coefficient of U between aqueous fluid and melt (K D = c fluid /c melt ) strongly increases with increasing HF content at 750 °C [25], as does the solubility of uraninite in uraninite-meltfluid systems (up to twenty times more U than in chloride solution with the same salinity at 770 °C [26]).These experimental studies demonstrate the role of complexing by fluoride for the hydrothermal transport of U in high-temperature (≥750 °C) magmatic hydrothermal systems.
Several studies have investigated the mechanisms by which hydrothermal fluids mobilize U to form high-grade U mineralization from low-to medium-temperature hydrothermal fluids (≤400 °C [22,27,28]).The current consensus is that oxidized fluids are favorable for U mobilization, while reduction is the key for U precipitation [27,[29][30][31][32].In this context, the role of F in controlling U transport remains poorly understood.According to the hard-soft acid-base theory, U(VI) and U(IV) are hard cations that form strong complexes with the hard ligand F − [33] but weaker complexes with the soft Cl − ligand.The stronger affinity of U for F is balanced by the fact that in natural fluids, Cl : F ratios are high (usually > ~100 [34,35]), because the solubility of F is limited by the solubility of minerals such as fluorite, fluorapatite, and F-bearing biotite.Hence, a quantitative approach is required to estimate the role of fluoride as a transporting agent for U in natural fluids.
The relative timing of U versus Cu mineralization at OD has received much attention recently.Macmillan et al. [15] divided OD uraninite into four groups based on textural and compositional characteristics.Class 1 uraninite has a  [15].Mineralogical and grade data are taken from Macmillan et al. [15].The RGB maps show Cu (blue), U (green), and Fe (red) distribution (colour scale at the bottom right).In (a) (uraninite-dominated ore), Cu and U distribution mirror each other, suggesting coprecipitation of the two metals.Mineralogy is dominated by uraninite.In (b) and (c), the main U minerals are coffinite and brannerite, and U distribution is diffuse and overprints the ore textures, suggesting a secondary origin.The maps were collected at the XFM beamline, Australian Synchrotron.See Li et al. [40] for details of data collection and processing.cubic-euhedral habit and relatively high Pb and ΣREY concentrations, as well as up to 7.42 wt.% ThO 2 ; this indicates precipitation from magmatic or high-temperature (>350 °C) hydrothermal fluids.In contrast, other uraninite classes show low ThO 2 contents, suggesting precipitation from lower-temperature (<250 °C) hydrothermal fluids [37].The late, massive, Class 4 uraninite has the lowest Pb and ΣREE + Y, and Macmillan et al. [15] suggest that it evolved from earlier-formed uraninites (Classes 1-3).X-ray maps of U and Cu distribution in ores from OD (Figure 1) illustrate the importance of secondary U mobility at OD.In a sample where most U is hosted in uraninite, the Cu and U distributions mirror each other, suggesting coprecipitation (Figure 1(a)).In contrast, U and Cu distributions differ markedly in a sample containing abundant coffinite; U mineralization is overprinting the ore textures, as well as occupying late fractures (Figure 1(b)).Similarly, in samples dominated by brannerite, U distribution appears to overprint early ore textures (Figure 1(c)); this is consistent with textural observations by Macmillan et al., [38] indicating a secondary origin for the majority of coffinite and brannerite.A protracted U mobility at OD is further supported by the U and Sm isotope study of Kirchenbaur et al. [39], which adds to geochronological evidence for gradual addition of U in several stages over 1000 Ma at elevated temperatures.Kirchenbaur et al.'s [39] data also indicate that U is sourced primarily from upper crustal (likely igneous) lithologies.
1.3.Aims.We calculated U(IV/VI) speciation in the simplified U-F-Cl-H 2 O-fO 2 system at 25-450 °C to compare the relative importance of F and Cl on U transport based on our current understanding of U thermodynamics.We have updated the thermodynamic properties of relevant U minerals and complexes, in particular U chloride and fluoride species compiled and reviewed by Bastrakov et al. [30] with the latest published data, and briefly discussed the choice of the key properties in our model.We then conducted reactive transport simulations to model the fluid-ore/rock interactions to help understand the hydrothermal mobilization of U and U-Fe-Cu-F association in IOCG systems, taking Olympic Dam as an example with well-described geological and mineralogical features.

Methods and Strategies for Thermodynamic Modelling
2.1.Mineral Solubility Calculations Using the Aliquot-Type Model.We use the aliquot-type equilibration model [44] to calculate the solubility of U minerals and the aqueous U speciation under simple P-T-x conditions for the H 2 O-U-NaCl-NaF-HCl-HF system (Figure 2(a)).In these models, one kilogram of acidic saline water with fixed Cl : F ratio of 100 (solution compositions, see Table 2) was reacted with uraninite (60 mmol) and U 3 O 8 (s) (20 mmol) to investigate the relative effects of fluoride and chloride.This Cl : F ratio reflects the composition of F-rich fluids in nature, which usually have Cl : F ratios of 100-200 [34,35].The salinity of the fluids was fixed at 15 wt.%NaCl equivalent (2.56 molal), which is similar, for example, to the hematite stage fluids at Olympic Dam [16,45].

2.2.
Step-Flow-Through Reactor Model.The aliquot equilibration model is not well suited to investigate the deposition and remobilization processes of U during fluid-rock interaction, because many reactions take place at different times and locations.; minerals are allowed to precipitate and dissolve, changing the composition of the rock and fluid at step 1.The equilibrated fluid is then extracted and allowed to react with the second compartment (step 2) and so on until the fluid exits the rock column.Repeating this process with fresh batches of fluids (waves 2 to N) allows to simulate increasing F/R ratio.
Hence, the number of waves represents the integrated effective F/R ratio, whereas the steps at each value of N represent a profile from the fluid source towards fresh rock (i.e., conditions buffered by fluids to conditions buffered by the rock).
In our model, we use an idealized composition for the protoores at Olympic Dam and a granite composition derived from published data of the Roxby Downs Granite (Table 3), which is the direct host for the ore-bearing hematite breccia at the Olympic Dam [36,45].The protoores consist of magnetite and contain low-grade U and Cu mineralization.The calculations were conducted at 200 °C, 3000 bar.The temperature reflects the lower end of the hematite stage at OD [16,45] and was selected because of the high uncertainties in the extrapolated logKs for critical U complexes at higher temperatures.Note that variation of pressure has little effect on the results in this temperature range, and test runs show that higher temperatures (to 300 °C) are unlikely to change the fundamental results of the simulations.

Selection of Thermodynamic Properties for Aqueous U Chloride and Fluoride Complexes, Uraninite, and Coffinite
Thermodynamic calculations were conducted using the HCh package developed by Shvarov and Bastrakov [46].Properties for minerals and aqueous species are mainly adopted from an updated version of the UNITHERM database [46,47], and we also included minerals and aqueous species (mainly Cu and Fe complexes) selected by R. Zhong et al. [48] and R. Zhong et al. [49].In particular, the thermodynamic properties of uranium minerals and aqueous species used for the calculations are listed in the supplementary material (Appendix, Table S1).Some plots were also drawn using Geochemist Workbench v.11 [50], with a customised thermodynamic database consistent with that of HCh.Here, we discuss the sources of properties for important U complexes and minerals.In general, we use the properties of U complexes selected by Bastrakov et al. [30], which are mainly based on the NEA (Nuclear Energy Agency) review of Guillaumont et al. [51].Noticeable exceptions include uranyl chloride complexes, where we use the new properties obtained recently by Migdisov et al. [52].We also use the one-term extrapolation method of Gu et al. [53] to propose new T-dependence for uranyl fluoride complexes based on the new knowledge of uranyl chloride complexes and U(VI) chlorides based on the knowledge of U(IV) fluoride complexes.Since the simulations indicated that some uranyl hydroxide complexes may be important as well, we also review the sources and reliability of these complexes.
3.1.U(IV) Aquo Complex.Despite the importance of lowoxidation state U in the nuclear energy industry, there is surprisingly little information about the nature and stability of U(IV) aquo complexes under hydrothermal conditions [30].As reviewed by Brugger et al. [54], U(IV) forms a spherical cation surrounded by 9-10 water molecules in the first coordination sphere in acidic solutions at room temperature [55][56][57][58].The thermodynamic properties of the U 4+ aqua ion are taken from Shock et al. [59].

U(IV) Fluoride
Complexes.The U(IV) fluoride complexes UF n 4−n with n = 1 -6 were included in this study.For the complexes with n = 1 -4, there is relatively good agreement at room temperature.Bastrakov et al. [30] selected the formation constants reported by Guillaumont et al. [51], which are the unweighted average values reported by Grenthe et al. [60], Noren [61], Kakihana and Ishiguro [62], Choppin and Unrein [63], and Sawant et al. [64].These values are also close to those selected by Grenthe et al. [65].In the absence of high-temperature data, Bastrakov et al. [30] used the van't Hoff equation to calculate the T-dependence of the complex to 100 °C and then the modified Ryzhenko-Bryzgalin (MRB) model [46] to extrapolate these properties to higher temperature and pressure.
The MRB parameters for UF 5 − and UF 6 2− are based on the free Gibbs energy of formation at 25-150 °C reported by Barsukov and Borisov [66].This provides a conservative estimate of the stability of high-order U(IV) fluoride complexes compared to the values selected by Guillaumont et al. [51].The formation constants for the reactions U 4 + + 5F − = UF 5 − and U 4 + + 6F − = UF 6 2− at 25 °C are 25.0 and 27.6, respectively, lower than the values calculated by Guillaumont et al. [51] based on the solubility data of Savage and Browne [67]  5 Geofluids temperature in acidic (pH 25 °C 0.24-1.61)HCl + LiCl solutions via in situ EXAFS spectroscopy.The average number of chloride increased from 1.0 ± 0.2 (Cl tot = 3 M) to 2.3 ± 0.4 (Cl tot = 6 M) and to 2.9 ± 0.5 (Cl tot = 9 M).We use the speciation information from these spectroscopic results as a benchmark to check the available thermodynamic properties for U(IV) chloride complexes.
The theoretical study of Shock et al. [69]  Bastrakov et al. [30] included just UCl 3+ and UCl 2 2+ in their model.The temperature dependence of the UCl 3+ complex was extrapolated using the entropy and enthalpy values selected by Guillaumont et al. [51], and the properties for UCl 2 2+ were taken from Barsukov and Borisov [66], who report free Gibbs energy of formation from the elements for UCl 2 2+ at 25-150 °C.Fitting to the MRB model provided extrapolation beyond 150 °C [30].However, including UCl 2 2+ resulted in unrealistically high predicted U concentrations, which led Bastrakov et al. [30] to ignore UCl 2 2+ for calculations at T > 200 °C.However, the UCl 2 2+ clearly needs to be included based on the average ligand number obtained from the study of Hennig et al. [68].
In order to attempt to obtain more realistic estimates for the stability of U(IV) complexes, we take advantage of the fact that U(IV)-F complexation is better constrained by experimental studies, which results in more reliable thermodynamic properties for U(IV)-F aqueous species at room temperature.We used the one-term isocoulombic extrapolation method developed by Gu et al. [53] to extrapolate thermodynamic properties for UCl 3+ and UCl 2 2+ to higher temperatures based on the properties of UF 3+ and UF 2 2+ .Then we fitted the properties with the OptimC program [70] and estimated new parameters of the MRB model for UCl 3+ and UCl 2 2+ .Note that the fitted logK data are based on the theory behind the isocoulombic approach [53] and that the MRB fitting is only a parameterisation of the logK data to be suitable for HCh database format.This is also the case for uranyl fluoride complexes discussed in Section 3.6.The newly fitted MRB parameters are listed in Table 4, and the extrapolated formation constants for UCl 3+ and UCl 2 2+ are compared with the extrapolations using Bastrakov et al.'s [30] parameters in Figure 3.The newly extrapolated formation constants (logK) for UCl 3+ are higher than those of Bastrakov et al. [30], while the logKs of UCl 2 2+ are lower than those from Bastrakov et al. [30] and Barsukov and Borisov [66].
We calculated the average numbers of chloride ligands using our thermodynamic model for the 3 M and 6 M Cl tot solutions analysed by Hennig et al. [68] to be 0.6 and 0.7, respectively.This indicates that the predicted chlorination numbers in our model are lower than the experimental ones (1.0(2) and 2.3( 4)) and that the model will tend to underestimate U concentrations in reduced brines.Although the properties of U(IV) chloride complexes at hydrothermal conditions clearly require more experimental evidence, the in situ XAS study of Hennig et al. [68] shows that UCl 2 2+ is important in chloride brines and needs to be included in the speciation model.The proposed extrapolation in this study appears to provide realistic estimates of U(IV) solubility in chloride brines.

Uranyl Oxycation (UO 2 2+
). Uranium (VI) is present in the form of the uranyl (UO 2 2+ ) ion, which forms a wide range of complexes with inorganic ligands such as hydroxide, halides, carbonate, or sulfate [54].In the absence of these ligands, the uranyl ion is hydrated by 5 to 6 water molecules, located on an equatorial plane relative to the two axial uranyl oxygens; the hydration number decreases with temperature, to ~4 at 300 °C, 250 bar [54].The thermodynamic properties of UO 2  2+ used in this study are compiled from Shock et al. [59], with HKF parameters based on regression of the experimental data of Hovey et al. [71] and Grenthe et al. [65].As reviewed by Shock et al. [69], the predicted U solubility values show good consistency up to 200 °C.However, it is difficult to evaluate the reliability of the thermodynamic properties at T > 200 °C due to the lack of experimental data.
The existence of UO , are especially important at high temperature (>250 °C).The recent in situ UV-Vis study of Migdisov et al. [52] was unable to confirm this result in relatively dilute (<1.5 M) NaCl solutions; the discrepancy could be due to the highly concentrated nature of LiCl solutions in the Dargent et al. [74] work or to deviations from the Beer-Lambert law caused by the wide range in ionic strengths investigated by Dargent et al. [74] (e.g., [75]).
Bastrakov et al. [30] calculated the logKs for reaction (1) (n = 1, 2) at 25, 50, 75, and 100 °C using the van't Hoff equation with ΔH r (enthalpy change of the reaction) and ΔS r (entropy change of the reaction) at 298.15 K recommended 6 Geofluids by NEA [51], and they fitted the calculated logKs to the MRB equation for high-temperature extrapolations.We compare the logKs of reaction (1) (n = 1, 2) from Migdisov et al.'s [52] UV-Vis study and MRB model, and the values calculated using the van't Hoff equation based on NEA [51] and the extrapolated values using Bastrakov et al.'s [30] MRB equation in Figure 4.The formation constants for UO 2 Cl + from Migdisov et al. [52] show good agreement with the values calculated from the van't Hoff equation and Bastrakov et al.'s [30] MRB at T < 100 °C.However, above 100 °C, the logKs from Migdisov et al. [52] become much higher than the values extrapolated using Bastrakov et al.'s [30] MRB model, indicating higher stability of UO 2 Cl + .For UO 2 Cl 2 (aq), the logKs from Migdisov et al. [52] are larger than the values calculated using the NEA values [30].Overall, Migdisov et al.'s [52] experimental study shows higher stability of UO 2 Cl + and UO 2 Cl 2 (aq) at elevated T (>100 °C) compared with recommendations from NEA [51] used by Bastrakov et al. [30].
In our model, we chose to include only the complexes that were characterised by Migdisov et al. [52], namely, UO 2 Cl + and UO 2 Cl 2 (aq).The UO 2 Cl 3 − complex was present only in small quantities in Migdisov et al.'s [52] experiments and is therefore not included in our model.[76].These values are in good agreement with data reported by Grenthe et al. [65].There is no experiment for uranyl fluoride complexes at elevated temperature.For UO 2 F + and UO 2 F 2 (aq), we extrapolated properties to 350 °C using the one-term extrapolation method of Gu et al. [53] based on the properties of UO 2 Cl + and UO 2 Cl 2 (aq) reported by Migdisov et al. [52].The new parameters of MRB model are listed in Table 4.
We compare the formation constants based on our new estimations with the values calculated from Bastrakov et al.'s [30] MRB parameters based on the NEA compilation [51] in Figure 3.The formation constants extrapolated by , we use the extrapolations proposed by Bastrakov et al. [30]; these two species do not appear to play a significant role in U transport in geochemical systems.
In Bastrakov et al.'s [30] compilation, the thermodynamic properties for UO 2 OH + and UO 2 (OH) 2 (aq) are expressed using the MHKF model based on Shock et al. [69] and NEA [51] data, while the properties of the polynuclear species are refitted with MRB equations based on the extrapolations of Plyasunov and Grenthe [80], which use the room-T thermodynamic data compiled by Grenthe et al. [65] and recommended by NEA [51].Note that due to the effect of entropy, the stability of polynuclear complexes decreases with increasing temperature [81].The recent experimental studies of Zanonato et al. [79] and Altmaier et al. [77] confirmed the reliability of the thermodynamic data for the polynuclear species recommended by NEA [51].Therefore, in this study, we use the MRB model fitted by Bastrakov et al. [30]  , and (UO 2 ) 4 (OH) 7 + .
Zanonato et al. [79] reported new hydrolysis constants, ΔH r and ΔS r for UO 2 OH + based on a temperaturedependent (10-85 °C) potentiometric study.Predictions based on these values (van't Hoff equation to 100 °C, fitted to MRB model) are compared to the NEA recommended values and to the extrapolations of Bastrakov et al. [30] in Figure 6(a).We also show the predictions based on the reviews of Nikolaeva [82] and Berto et al. [78].In general, the new data of Zanonato et al. [79] are close to the NEA-selected values and those of Nikolaeva [82].Berto et al. [78] underestimates the formation constant and is excluded from this discussion.We selected the new extrapolation based on the latest data of Zanonato et al. [79] for our calculations (Table 4).
The thermodynamic properties for crystalline U oxides (listed in Appendix, Table S1) are taken from Guillaumont  [52] experimental study and their fitted MRB models, values calculated using van't Hoff equation using ΔH r and ΔS r recommended by NEA [51] and predicted values from Bastrakov et al.'s [30] MRB model.8 Geofluids et al. [51].The C p parameters are collected from Grenthe et al. [65] and are valid at 250-600 K.These values are close to those reported by Fink [85] so that within the temperature range of our calculations, these values are reliable.9 Geofluids stable with changing temperature, while the concentration of the free fluoride ion (F − ) decreases quickly with increasing temperature at the expense of forming the HF(aq) ion pair.With increasing temperature, the stability of U(VI)-F complexes decreases, while U(VI)O 2 Cl + and U(VI)O 2 Cl 2 (aq) become predominant above 200 °C.In contrast to the reduced system (Figure 7(a)), the solubility of U 3 O 8 (s) remains relatively high with increasing temperatures.

Results
An important feature in both U(IV)-and U(VI)dominated aqueous systems is that fluoride complexes are stable at low temperatures and chloride complexes at high temperatures.This is further illustrated in a temperature versus log fO 2 (g) diagram (Figure 8(a)) and is a result of the strong association of the hydrofluoric acid (HF(aq) ion pair) at elevated temperature compared to HCl(aq). Figure 8(b) shows that low pH favors uranyl chloride over fluoride species.

Mobilization and Precipitation of Uranium with
Increasing Fluid: Rock Ratio in Magnetite-Rich Protoores.Results of the step-flow reactor model are shown in Figures 9 and 10.The dominant feature is the development of a sharp reaction front as a result of the replacement of magnetite in the protoore by hematite.Uranium and Cu concentrations are elevated in the ore at the hematite-magnetite transformation boundary (Figures 9(a) and 9(b)).Similar results were obtained by Bastrakov et al. [45] for Cu and Au enrichment at the hematite-magnetite reaction front and indicated that the oxidizing fluids remobilize redoxsensitive elements and (re-)precipitate them due to the changes in redox and pH conditions at the reaction front.A key result from our model is that fluorine is enriched in the form of fluorite at the reaction front.Fluorite concentrates together with Cu within the hematite zone ahead of the reaction front.
With increasing F/R ratio, the reaction front moves further into the protoore (Figures 9(a .This is accompanied by the continuous upgrading of U, Cu, and F near the reaction front (Table 5).The calculated mineralogy of the ores at two F/R ratios is listed in Table 5.For a F/R ratio of 10 : 1, the uraninite concentration at the reaction front reaches 0.6 wt.%, which is 150 times higher than the weakly reacted ores (0.004 wt.% uraninite + U 3 O 8 (s)).At F/R = 15, the uraninite content at the reaction front increased to 0.92 wt.%.The model predicts that UO 2 OH + is the predominant aqueous species; significant amounts of U are also carried as uranyl fluoride and chloride complexes (Figures 9(c) and 9(d)).Most U precipitated at the reaction front due to the destabilization of uranyl species caused by changes in pH and log fO 2 (Figure 10).The predominant sulfur species are SO 4 2− and HSO 4 − at the reaction front, but their concentration drops quickly towards the magnetite side.In contrast, HS − and H 2 S(aq) become more important across the reaction front (Figures 10(a The main Cu minerals at/near the reaction front are bornite (±chalcocite) and chalcopyrite, with bornite usually more abundant than chalcopyrite (Table 6).Uranium is predicted to precipitate as uraninite.The simulations show the 10 Geofluids following mineral zonation at the reaction front: chalcopyrite and fluorite are usually enriched towards the magnetite side, while bornite (±chalcocite) and uraninite are towards hematite ore side (Figure 9).From these results, the predominant reaction is the oxidation of magnetite, which is coupled with the reduction of sulfate (and oxygen) and an increase in pH: The oxidized fluids (low reduced S contents) cause the transformation of chalcopyrite into Cu sulfides with higher Cu : S ratios, for example, The redox front associated with the reduction of sulfate promotes the reduction of U as well; the deposition of uraninite can be expressed for example as   2 for system composition).(a) Solubility of UO 2 (s) under reduced conditions.Below ~260 °C, U(IV)-F complexes predominate over U(IV)-Cl complexes.The overall solubility of UO 2 (s) decreases with increasing temperature.(b) Speciation of the chloride and fluoride ligands for the simulation in (a).(c) Solubility of U 3 O 8 (s).The predominant species below ~170 °C are U(VI)O 2 F 2 (aq) and U(VI)O 2 F + ; uranyl chloride complexes become predominant at T > 170 °C.These calculations were performed in a closed system, with pH and redox self-buffered.The system contains 60 mmolal U, and calculations are isobaric at 3 kbar.
These geochemical constraints are consistent with the textural observations of Macmillan et al. [38].Brannerite similarly appears to be a late mineral, forming at least in part via interaction of U-bearing fluids with preexisting Ti minerals [38].Hence, precipitation of coffinite and brannerite versus uraninite are controlled by temperature and by the presence of suitable microenvironments [87].Our equilibrium model does not account for these microenvironments, but the results would be very similar if coffinite/brannerite were to precipitate rather than uraninite.

Fluorite Enrichment with Cu-U Mineralization. A correlation between U and F enrichment has been found
in the OD ores [15,36,38].The U-F concentrations for Cu-unmineralized (Cu < 300 ppm), weakly mineralized (Cu = 300-1000 ppm), moderately mineralized (Cu = 1000-3000 ppm), and mineralized OD samples (Cu ≥ 3000 ppm) are plotted in Figure 12.Independently from Cu grade, U concentrations are generally positively correlated with F content, except for ores containing >60 wt.% hematite where fluorite concentrations drop down but U 3 O 8 concentrations maintain an upward trend, especially for samples containing ≥1000 ppm Cu (Figures 12(c) and 12(d)).Macmillan et al. [15] reported that the highest U grades are often contributed by massive uraninite found in hematite breccia, and these uraninites are closely intergrown with Cu sulfides.
Our calculations show that U and Cu enrichment at the hematite-magnetite reaction front is associated with F enrichment (Figure 9; Table 5).Fluorine concentration in the hematite ores also increases with continuous hydrothermal alteration (Table 5).It is important to note that these enrichments result from interactions with fluids that are not particularly F-rich: the starting composition of fluid in each wave is in equilibrium with granite, and its F content is relatively low (F = ~80 ppm, Cl : F = ~1800) compared to natural F-rich fluids that can contain in excess of 300 ppm F (magmatic-derived fluids at the Capitan Pluton [34,35]).In the step-flow reactor model, we assume there is no F in the protoores and the granite is the only unit that contributes F (Table 3).Therefore, F in the hematite ores originates from the granite and is transported by the fluids.For each wave of fluid, the F concentration dropped by about 9.5 ppm (12% of the total F in the fluid) at the hematite-magnetite reaction front, indicating that F kept being precipitated at the reaction front (up to 0.09 wt.% for a F/R ratio of 15).In general, the predicted F grades are lower than the observed grades.This reflects the fact that the protoore in our model does not contain F or Ca.Higher F grades would result in the protoore containing Ca in particular.
A typical textural relationship for uraninite, bornite, and chalcocite is illustrated in Figure 13 based on the observations of Macmillan et al. [15].Bornite and chalcocite are more likely to be precipitated at the same stage while fluorite is overprinted by uraninite, bornite, and chalcocite (Figure 13).The formation of this texture indicates that fluorite formed earlier than uraninite and Cu sulfides; this is consistent with our simulations (Figure 9).The processes to form this bornite-uraninite-fluorite association are further illustrated in Figure 14.According to the model, fluorite precipitates several steps earlier than uraninite and Cu sulfides along the flow direction of fluids.This results in the local zonation from fluorite to uraninite and Cu sulfides upon increasing F/R ratio, as early formed fluorite is overprinted by the later formed uraninitebornite ± chalcocite assemblage with the influx of new fluids.Hence, the F-U base metal association in this case does not reflect an active role of F as a transporting and/or precipitating ligand but rather results from conditions at the reaction front affecting F (via pH change), U, and Cu (pH and redox changes).Therefore, the association of F and U-Cu enrichment may simply be formed by the coprecipitation of fluorite due to the similar local physical-chemical traps rather than the breakdown of U-F complexes, but it does not reflect the F-rich nature of the ore-bearing fluids, since the calculated F concentration in fluids indicates only low-F fluids, and the predominant U species is a uranyl hydroxyl complexes (UO 2 OH + ) (Figures 9(c  13 Geofluids aqueous complexation only at relatively low temperature (50-200 °C) while at elevated temperature (>200 °C), the relative stability of U(IV/VI)-F complexes drops dramatically and U(IV/VI)-Cl complexes become more important in fluids (Figures 7 and 8).Uranyl hydroxide complexes also appear to be important even slightly acidic pH associated with silicate-buffered fluids; our simulations suggest that a mixture of hydroxide, chloride, and fluoride complexes transported U at OD (Figures 9 and 10).
As a hard ligand, fluoride forms strong complexes with cations such as Fe 3+ , U(VI)O 2 2+ , and REE 3+ , which makes it a promising ligand for transporting these metals   14 Geofluids [2,33,[88][89][90].Consequently, it has long been proposed that F may play a key role for U transport in hydrothermal fluids [2,25].Comparing F and Cl, our calculations show that the fluoride complexes of both U(IV) and U(VI) dominate in fluids only at relatively low temperatures (<~200 °C).At elevated temperatures (>200 °C), U-F complexation is not important.When T is above 200 °C, uranyl chloride complexes dominate in acidic oxidized fluids (Figures 7 and 8).As illustrated in Figure 15, U(IV/VI) fluoride complexes are much stronger than the corresponding chloride complexes; the difference in formation constants decreases with increasing temperature but remains significant at 300 °C.The increasing importance of chloride versus fluoride complexes hence reflects the stability of the HF(aq) ion pair at high temperature, which competes with the formation of fluoride complexes.Another important factor is that the availability of the fluoride ion in ore fluids is very much limited by extremely low solubility of many fluoride minerals.For example, the solubility of fluorite (CaF 2 ) is very low (K sp = 6 6 × 10 −8 at 200 °C), but CaCl 2 is very soluble, thus causing the high Cl : F ratio in the fluid.In summary, the transport of U by fluoride may only be of significance in fluids with T < 200 °C while at elevated temperature, U-F complexes are not stable and contribute little to U transport.

Remobilization and Upgrading of Uranium by
Hydrothermal Fluids.At OD, large-scale hydrothermal activity was responsible for ore formation, and the hydrothermal history of the deposit was complex and protracted [36,39,[91][92][93].As discussed before, our calculations show that U is highly enriched at the magnetite-hematite transformation front (Figure 9).The relative enrichment of U depends on the F/R ratio.According to our calculations, the major U and Cu minerals at the reaction front are uraninite, chalcopyrite, bornite, and chalcocite (Table 5; Figure 9).The relative enrichment of U compared to protoore (U reaction front /U protoore ) is 150 at F/R = 10 and 225 at F/R = 15.Similarly, the Cu contents increase rapidly with increasing F/R, resulting in a zoning of chalcopyrite-bornite (±chalcocite) around the reaction front (Figure 9).In detail, the model predicts that bornite (±chalcocite) is usually accompanied by uraninite enrichment, while chalcopyrite is accompanied by fluorite enrichment.Li et al. [94] demonstrated experimentally that reactions among sulfide minerals create microenvironments that result in efficient scavenging of U from solution and provide evidence that kinetic factors also favor the enrichment predicted by the equilibrium thermodynamic calculations [95].
Results from the step-flow reactor model are consistent with mineralogical observations from OD. Deposit-scale bornite-chalcopyrite zonation is a well-established feature of the deposit, and the highest Cu mineralization is also found along this boundary [91,96].Uraninite is often locally enriched near the chalcopyrite-bornite interface, accompanied by fluorite enrichment [16].Therefore, continuous alteration by oxidized fluid will cause the local dissolution and reprecipitation of the U-Cu ores, which keeps upgrading the U-Cu grade around the bornite-chalcopyrite interface, at deposit but also at the local scale (Figures 9 and 16).

Sourcing and Deposition of U by Hydrothermal Fluids.
The sourcing processes of U to form deposits have long been  15 Geofluids a controversial topic [2,39].The relationship between U solubility, fluid salinity, and pH has been investigated by Richard et al. [32] on natural fluid inclusions and with experiments.Their results show that acidic (pH = 2 5 -4 5) saline solution caused more U to be dissolved, indicating that chloride and acidic conditions are important for U transport.However, as indicated by our calculations, fluoride is important for U mobilization only at relatively low temperatures (<250 °C).Hence, the F-rich nature of Olympic Dam and other U deposits may not necessarily reflect a F-rich nature of the ore-forming fluids but more likely reflects the nature of source rocks or circulation pathways of fluids.This is further supported by the case of the Oak Dam IOCG deposit, where F is absent in the Cu-U ores [97], as a result of the absence of a F-rich igneous unit in the region.
An interesting outcome of the model is the importance of the U(IV) complexes at high temperature under reduced conditions (Figure 7(a)).As discussed in Section 4 and highlighted by Bastrakov et al. [30], confirming these results requires new experimental data, although the room temperature properties are well established.Aside from confirming the stability of U(IV) halide complexes at elevated T, experimental studies need to assess the role of mixed halidehydroxide-aquo complexes at elevated P-T.Ferri et al. [76] found no evidence for such ternary complexes at room 16 Geofluids temperature.However, in the case of Zr(IV), Migdisov et al. [98] showed that 8-or 6-coordinated fluoro-aqua complexes predominate at low temperature and tetrahedral hydroxyl-fluoro complexes predominate at high temperature.Such complexes could further increase U(IV) mobility at elevated temperature.
In conclusion, the dissolution and precipitation of U are controlled by T-pH-ƒO 2 , as well as the ligand type The precipitation of uraninite via reduction of uranyl complexes can proceed with bisulfide (see ( 4)) or magnetite (Fe 2+ ) as reductants.
Accordingly, increased ƒO 2 and [H] + will enhance uraninite dissolution, while a decrease will result in uraninite precipitation (Figure 8(b), U precipitates towards higher pH).Richard et al. [32] postulated that acidic brines with pH between 2.5 and 4.5 are required to maintain high U concentrations in basinal fluids.In addition, the ligand type is also important for U dissolution.By forming strong complexes with [UO 2 ] 2+ , hard ligands such as F − , CO 3 2+ , and SO 4 2− may be essential in helping uraninite dissolution.Also note that because of the strong association of F at low temperature, F is likely to be significant as a ligand only at T < 200 °C.

Final Word: the Geochemistry of IOCG Deposits
IOCG deposits are an increasingly important source of economic Fe, Cu, Au, and U since the discovery of the giant Olympic Dam Cu-U-Au deposit in South Australia in 1975 [99].What constitutes an IOCG deposit has been initially formulated based on the Olympic Dam discovery; that is, IOCG had no formal genetic meaning, and the formal  Generally, the attempt of Groves et al. [99] is considered the most authoritative.Our calculations show that as a class, the common geochemical features of IOCG deposits as described by Groves et al. [99] can be related to the two-phase process we modelled, whereby a magnetite-hematite-rich orebody (formed via a number of processes under different tectonic settings) is enriched in Cu ± U,F through a second stage of hydrothermal circulation.Groves et al. [99] outline the following features as being necessary for an IOCG sensu stricto deposit such as Olympic Dam: (i) Cu and Au as economic metals, (ii) hydrothermal features and structural controls (e.g., breccias), (iii) abundant iron oxides (hematite, magnetite), (iv) LREE enrichment and low S sulfides like chalcopyrite and bornite, and (v) lack of abundant syn-sulfide quartz veins.Groves et al. [99] also note that the formation of OD-style IOCGs involved mixing of crustal and meteoric fluids with higher temperature magmatic/metamorphic fluids.In our model, the magmatic stage is not modelled but can correspond to the formation of the magnetite-rich protoore.Furthermore, a temporal (not spatial) relationship with magmatism is a key feature of IOCG deposits according to Groves et al. [99].This feature is outside the scope of our geochemical model, although it can relate to the heat source driving hydrothermal fluids or to a geodynamic setting conductive to large fluid flow.
Our modelling also indicates that the coenrichment of F and U in IOCG ores most likely reflects the source of the ore-forming fluids and does not reflect an essential role of F in controlling the metal endowment of these deposits.

Figure 1 :
Figure1: Contrasting U-Cu distribution in three samples from the Olympic Dam IOCG deposit, South Australia, illustrated by synchrotronmicro-XRF imaging of selected samples described in detail by Macmillan et al.[15].Mineralogical and grade data are taken from Macmillan et al.[15].The RGB maps show Cu (blue), U (green), and Fe (red) distribution (colour scale at the bottom right).In (a) (uraninite-dominated ore), Cu and U distribution mirror each other, suggesting coprecipitation of the two metals.Mineralogy is dominated by uraninite.In (b) and (c), the main U minerals are coffinite and brannerite, and U distribution is diffuse and overprints the ore textures, suggesting a secondary origin.The maps were collected at the XFM beamline, Australian Synchrotron.See Li et al.[40] for details of data collection and processing.

3. 5 .
Uranyl Chloride Complexes.The formation of UO 2 Cl n 2 − n complexes is described by the following equation:

4. 1 .
Comparison of U-F and U-Cl Complexation.The effects of Cl and F on U complexation were compared by calculating the speciation of U in solutions containing up to 60 mmol U present as uraninite (Figure 7(a)) or U 3 O 8 (s) (Figure 7(c)).Figure 7(a) shows the speciation of U(IV) aqueous species in reduced fluids.At T < 170 °C, U(IV)F 4 (aq) is the predominant species while U(IV)F 3 + and U(IV)F 2 2+ are less important.Above 170 °C, the concentrations of U-F species decrease quickly and U(IV)Cl 2 2+ becomes predominant at T > 260 °C.As shown in Figure 7(b), the free chloride ion (Cl − ) dominates and its concentration remains relatively

Figure 5 :
Figure 5: Uranium solubility in 0.51 molal NaCl solutions in equilibrium with metaschoepite or Na 2 U 2 O 7 •H 2 O(cr) at 298.15 K plotted as a function of pH.(a) Uranium solubility calculated with the thermodynamic model used in this study, compared to the solubility data of Altmaier et al. [77].(b) Distribution of species for the simulation in (a); predominant species are highlighted.
) and 9(c) versus Figures 9(b) and 9(d)) ) and 10(b)).The pH decreases at the reaction front and increases dramatically as the fluid equilibrates with the protoore (Figures10(c) and 10(d)); this is accompanied by a dramatic decrease in the oxygen fugacity of the fluids (Figures 10(e) and 10(f)).

Figure 7 :
Figure 7: Calculation of U solubility and speciation in acidic F − -and Cl − -bearing solutions at 25-450 °C (see Table2for system composition).(a) Solubility of UO 2 (s) under reduced conditions.Below ~260 °C, U(IV)-F complexes predominate over U(IV)-Cl complexes.The overall solubility of UO 2 (s) decreases with increasing temperature.(b) Speciation of the chloride and fluoride ligands for the simulation in (a).(c) Solubility of U 3 O 8 (s).The predominant species below ~170 °C are U(VI)O 2 F 2 (aq) and U(VI)O 2 F + ; uranyl chloride complexes become predominant at T > 170 °C.These calculations were performed in a closed system, with pH and redox self-buffered.The system contains 60 mmolal U, and calculations are isobaric at 3 kbar.

Figure 8 :
Figure 8: Mineral solubility and speciation for the U-F-Cl system.Minerals are labelled in bold; aqueous predominance fields are shown in the mineral solubility fields with thin red lines.(a) Diagram showing uraninite solubility and U speciation as a function of T ( °C) and ƒO 2 .Uraninite is poorly soluble at the phase boundary of magnetite and hematite, and at T < ~170 °C, U(VI) fluoride complexes predominate while at T > 170 °C, U(VI)O 2 Cl 2 (aq) is predominant.(b) Uraninite solubility and U speciation as a function of pH and O 2 (aq) at 200 °C.
) and 9(d)), with contributions from chloride and fluoride complexes.

5. 2 .Figure 9 :
Figure 9: Results of the step-flow-through reactor model.(a-b) Ore mineralogy after 500 waves and 1000 waves, which represent the fluid/ rock ratio at 10 : 1 and 20 : 1, respectively.The direction of the reaction front of magnetite replaced by hematite is indicated by the arrows in (a) and (b).Uraninites and Cu sulfides (Ccp and Bn) are precipitated at the reaction front, accompanied by fluorite coprecipitation.Fluorites are enriched in the fully reacted ores (hematite > 95%).Zonation is found at the reaction front, with chalcopyrite and fluorite enriched near the magnetite side and bornite (±chalcocite) and uraninite enriched at the hematite side.(c-d) Main aqueous U species.

Figure 10 :
Figure 10: Main sulfur species, pH, and ƒO 2 in the solution for results from the step-flow-through reactor model at F/R = 10 (a, c, e) and 15 (b, d, f).

Figure 11 :
Figure 11: Stability of coffinite versus uraninite as a function of silica activity and temperature compared to the solubility of quartz and amorphous silica.
(F, Cl, and hydroxide are investigated in this study).The dissolution of uraninite can be described by the following equation: U IV O 2 uraninite + 0 5O 2 + 2H + + x F, Cl, OH − = U VI O 2 F, Cl, OH x 2−x + H 2 O 5

Figure 14 :Figure 15 :
Figure 14: Schematic model for the formation of uraninite-bornite-fluorite intergrowth textures formed at the reaction front according to the step-flow-through reactor calculations.(a) Mineralogy at the reaction front after N waves; fluorites are precipitated a bit earlier than uraninite and Cu sulfides, forming mineralogical zonation; (b) mineralogy at the reaction front after N + k waves; the reaction front is moved towards the fluid penetration direction, causing U-Cu and fluorite mineralization to be moved in the same direction.Due to this movement, early formed fluorite will be overprinted by uraninite and bornite (and/or chalcocite) when fluorites are not fully mobilized.

Figure 16 :
Figure 16: Predicted maximum U, Cu, and F contents with F/R ratio.

Table 2 :
Simulated system for U-Cl-F-H 2 O equilibrium.

Table 3 :
System composition used in the step-flow reactor model.
UCl 4 (aq), although HKF parameters or property sources are not listed.Shock et al.'s [69] model predicts high stability of U(IV) chloride complexes above 100 °C at low pH; for example, UCl 2 2+ can account for up to 8% dissolved U in the solution with a pH below 2, and UCl 3 +and UCl 4 (aq) dominate above 250 °C at pH = 1.

Table 4 :
Parameters of the MRB model for U(IV/VI) aqueous species fitted in the present study.

Table 5 :
Ore mineralogy after different waves.

Table 6 :
Maximum U, Cu, and F at reaction front.